Fraudulent ICO has learned to identify with the help of artificial intelligence
Scientists from the University of California, Michigan and Stanford have created an algorithm that is based on deep machine learning is able to detect fraudulent ICO up to 83%. MarkLog wrote about this with reference to the joint communication of the universities.
The researchers downloaded data on 2251 ICO were Nacala 2016. The algorithm has apostoles the lifetime of the project and the change in the price of the cryptocurrency with information about the ICO (whitepaper, information about the founders, the sate of the project, social networks and other data).
The study says that unearthing will allow investors to identify fraudulent ICOS at an early stage and save their money. "We hope that this work will help investors to identify scammers and attract more efforts for the automatic evaluation and analysis of ICO projects," the scientists write
Scientists from the University of California, Michigan and Stanford have created an algorithm that is based on deep machine learning is able to detect fraudulent ICO up to 83%. MarkLog wrote about this with reference to the joint communication of the universities.
The researchers downloaded data on 2251 ICO were Nacala 2016. The algorithm has apostoles the lifetime of the project and the change in the price of the cryptocurrency with information about the ICO (whitepaper, information about the founders, the sate of the project, social networks and other data).
The study says that unearthing will allow investors to identify fraudulent ICOS at an early stage and save their money. "We hope that this work will help investors to identify scammers and attract more efforts for the automatic evaluation and analysis of ICO projects," the scientists write
This was pretty weird. I meant about the total coins in the coinmarketcap was no more than 1.6k but it says more than 2.2k.
But is the algorithm already tested? How many accuracies have been getting the algorithm itself?
This is a system and it can be failed anytime. I meant it doesn't like a good thing for me.
Some scam projects are using fake identities and how they are learning it? I know some scam projects were taking the fake identity from videos on youtube.
AI still hasn't come to the state when it can resolve such complicated tasks like this where it needs to find the scammy ICO.
Scientists can brag and show the results but they are just graphs which can pull the investors to invest into recommended ICOs.
I don't know if the OP thought about the possibility that this is just a phishing trap to invest in other ICOs that were represented as positive.
What if this smaller percent of the positive and truthful ICO is actually revealed to be a scammy project.
Even with the algorithms, the human touch needs to be put for the necessary research of the possible scam and not one AI can compete yet with the human's power to distinguish the scam from the proper project.
